[16] Carbohydrate sequence analysis by electrospray ionization-mass spectrometry

[161

CARBOHYDRATE SEQUENCE ANALYSIS BY E S - M S

377

Future Developments Significant instrumental improvements in both cost and performance are to be expected as several types of mass spectrometers reach a level of refinement that permits commercial versions to be introduced. Ion trap mass spectrometers are increasingly used for the analysis of biomolecules with greater sensitivity and mass resolution than quadrupole mass spectrometers.35 37 Sequence information for peptides can now be obtained on reflectron time-of-flight mass spectrometers and improvements in the ability to scan the reflectron voltages are expected. > Improvements in surfaceinduced dissociation (SID) have been achieved by coating metal surfaces with self-assembled monolayers, 3s-39greatly increasing the efficiency of fragment ion recovery from the surface. In addition, fragment ions have been observed in SID studies of peptides that allow the differentiation of leucine and isoleucine on quadrupole-type mass spectrometers, a° These developments will ensure that mass spectrometry continues to have a significant role in biological research. .~5 K. A. Cox. J. D, Williams, R. G. Cooks, and R. E. Kaiser, Jr., Biol. Mass Spectrom. 21, 226, 1992. ~'+J. A. Castoro and C. L. Wilkins, Anal, Chem. 65, 2621 (1993). 37 M. W. Senko, S. C. Beu, and F. W. McLafferty, A n a l Chem. 66, 415 (1994). 3s B. E. Winger, R. K. Julian, R. G. Cooks, and C. D. Chidsey. J. Am. C,~enT. Soc. 113, 8967 (1991). 3+ V. H. Wysocki, J. L. Jones, and J. M. Ding, ,I. Am. (,+hem. Soc. 113, 8969 (1991). 4o A. L. McCormack, A. Somogyi, A. R. Dongre, and V. H. Wysocki. Anal. ('hem. 65, 2859 {1993).

[1 61 C a r b o h y d r a t e Sequence Ionization-Mass

By VERNON N.

REINHOLD,

Analysis by Electrospray Spectrometry

BRUCE B.

REINHOLD,

and STEPHEN CHAN

Introduction Carbohydrate materials with a diversity of structural types and acid lability pose unique analytical problems. A treatise in this series has been totally devoted to these endeavorsY This chapter may serve as a bridge between this volume and an earlier volume on mass spectrometry. 2 This i G. W. Hart and W. J. L e n n a r z (eds.), Methods Enzymol., 230, (1993). + J. A. McClosky (ed.), Mass spectrometry. In Methods Enzymol. 193, (1992).

METttOI)S IN ENZYMOI+OGY,VOL. 271

Copyright c~, 199(+by Academic Press. inc. All rightsof reproduction in any formreserved.

378

MASS SPI~CTROME I'RY

[161

strong focus to understand carbohydrate structure becomes most germane with the growing interest in the functional roles attributed to these moieties. As researchers sharpen their focus on the biochemical basis of cell-cell interaction, carbohydrates and their glycoconjugates are often found to be crucial participants. The evidence for these interactions has been growing and is strongly supported by the use of carbohydrase inhibitors, deglycosylating enzymes, and site-directed mutagenicity of participating glycans. From these studies carbohydrates appear to serve as the " V e l c r o " of adhesion, the decoys against bacterial invasion, the modulators of protein structure, and epitopes for molecular targeting and trafficking. Numerous human parasites, through molecular mimicry, have capitalized on these intricate functional roles to subvert immune surveillance, invade, and thrive within human cells. In nitrogen fixation, bacterial oligosaccharides play major roles in their relationships with plants. Starting from recognition and attachment, and concluding with a new plant organ, the root nodule, carbohydrate involvement in these symbiotic events is most specific and pronounced. 3"4 Cell-specific carbohydrate structures are known to regulate events during differentiation: and the same sites serve as microbe attachment and invasion sites. 6 Receptors on lymphocytes that mediate adherence have been found to be a small family of glycosylated molecules, the selectins. 7 Wherever confining m e m b r a n e surfaces occur, and signal transduction is carried out, the diversity and specificity of the carbohydrate molecule are likely to be utilized as the conduit for this communication. Although a precise understanding of function is frequently compromised by structural complexity, the overall conclusions remain that these residues are major participants in the fine tuning of cellular processes. As in many areas of science, revelation will come with an understanding of detail. From a structural standpoint, oligosaccharides are ideally suited for events requiring molecular specificity, e.g., their polyhedric character provides a platform with numerous functional groups for modification or interaction; m o n o m e r s within a chain can participate in multiple linkage and branching patterns, creating arrays of unique possibilities. It is for these very reasons that methodology for carbohydrate sequencing remains undeveloped. With the absence of optical properties for detection or specific functional groups for modification, and their copious structural isomerism,

L. J. Halverson and G. Stacey, Microbiol. Rev. 53, 193 (1986). 4 H. Spaink. D. Sheclcy, A. van Brussel, J. Glusha, W. York, T. "lak. O. Geigcr, E. Kennedy, V. N. Reinhold. and B. Lugtenberg, Nature (l,ond(m) 354, 125 (1991). " Z. Zhu. L. Chcng, Z. Tsui, S. Hakomori, and B. A. Fenderson..l. Reprod. Fertil. 95, 813 (1992). ~ K.-A. Karlsson, A n m c Rev. Biochem. 58, 309 (1989). 7 L. A. Lasky, Science 258, 964 (1992).

[16]

CARBOHYDRATE SEQUENCEANALYSISBY ES-MS

379

carbohydrates remain the analytical challenge of contemporary biology. Pursuit of linkage and branching detail is a major undertaking and rarely are characterizations complete. Methylation analysis, and the identification of methylated alditol acetates, fails in principle to identify all entities uniquely, and this information cannot be integrated into a sequential array. Moreover, methylation analysis requires large quantities of material compared with the typical operational range of a mass spectrometer. Sequence has a somewhat more entangled meaning for a variably linked and multiply branched structure than for the linear biopolymers of amino and nucleic acids. Microheterogeneity, a consequence of variable chain termination, adds further complication, and these features are compounded by the stereochemistry of glycosidic bonds (anomers) and isobaric residues. We can take some satisfaction in the knowledge that sequence determination reduces to defining the type of motif (glycotype), its distribution (glycoform), and a precise linkage and branching pattern (glycomer: the term glycomer is an extension of earlier notation, glycoform and glycotype, TM defining the detailed structure of a glycan). This diversity of function, the fidelity of biosynthesis at any one site or from any one cell, represent a small part of the many intriguing questions in glycobiology.

Electrospray Ionization-Mass Spectrometry The major theme of this chapter relates to a detailed characterization of carbohydrates with component applications focused on glycoprotein glycans. A previous volume covered the broad area of mass spectrometry with selected chapters on electrospray ionization." This chapter introduces and discusses electropray ionization-mass spectrometry (ES-MS) as applied to the sequence analysis of glycoprotein oligosaccharides. Here, we describe the instrumental power of mass separation and collision-induced dissociation (CID) to arrive at detailed carbohydrate structure. Developments in generating ions from a condensed phase have brought new dimensions to our understanding of biopolymers. Undertakings that were inconceivable a few short years ago are now routine, and much of that credit goes to the desolvation technique of electrospray s (ES) ionization. As applied to biopolymers, the first established success of ES was realized with polar molecules ionized from aqueous aerosols, a development highly suited to peptide sequencing. In contrast to this important focus, this chapter emphasizes the value of ES when applied to lipophilic samples evaporated 7~T. W. Rademacher. R. B. Parekh. and R. A. Dwek. Ann. Rm. Biochem. 57, 785 (1988). J. B. Fcnn, M. Mann, C. K. Mcng, S. F. Wong. and C. M. Whilehouse. Science 246, 64 (1989).

380

MASSSPECTROMETRY

[ 16]

from nonpolar solvents, a strategy concordant with the detailed analysis of oligosaccharide linkage and branching. Mass spectrometry of carbohydrates has been largely based on ballistic ionization strategies, e.g., fast atom bombardment (FAB) MS, which exhibits poor ionization efficiency and allows considerable ion source fragmentation. For the purposes of carbohydrate analysis, ES is the most effective and generally applicable technique currently available for transforming these molecules into gas-phase ions. The technique makes use of high electric fields to aerosolize solutions, creating a fine mist with each droplet carrying an excess surface charge. A heated, countercurrent bath gas at ambient pressure evaporates neutral solvent from the droplet surface, reducing its size and hence increasing the surface charge density. Electrostatic repulsion disrupts the surface, leading to the ejection of smaller charged droplets with the eventual development of gas-phase ions. In operation, solutions (usually between 1 and 20/xl/min) are directly infused through a stainless steel needle maintained at a few kilovolts relative to the chamber walls, which induces the surface charging of the emerging liquid and aerosol formation. The droplets migrate toward a capillary inlet through the countercurrent bath gas, which hastens solvent evaporation and serves to isolate the high-vacuum analyzer from atmospheric pressures. Ions that enter the glass capillary emerge as a supersonic free jet and pass through skimmers into the analyzer (Scheme I). Electrospray provides two features important for biopolymer characterization: efficient ionization and multiple charging, and both factors contribute to successful CID spectra. The former characteristic produces intense ion beams while the latter allows higher molecular weight determinations. Electrospray spectra of a single protein provides an envelope of ions with the degree and distribution of charge a property of the protein and its milieu. For oligosaccharides, additional ion envelopes ensue as a consequence of microheterogeneity, where any one envelope specifies a glycoform distfibu-

Electrodes

Ions

Ptmlping Zones S(HEME [

[16]

CARBOHYDRATE SEQUENCEANALYSISBY ES-MS

381

tion. Although the combination of multiple charge states and microheterogeneity produces a complex spectrum, in practice they can be easily deconvoluted to yield data from which molecular weights are extracted. Treatment of such data in specific ways suppresses artifacts and improves spectral dynamic range and confidence in peak detection9 We illustrate several applications in an effort to arrive at a global strategy for oligosaccharide sequencing.

Carbohydrate Glycan Profiling lntrodttction

The structural characterization of an oligosaccharide involves gathering molecular detail on three levels: the compositional makeup, the biopolymer array or branching pattern, and an understanding of interresidue linkage. In general, similar elements of structure comprise the glycans of glycoproteins, which have been referred to as glycotype, glycoform, and glycomer. Carbohydrate chains on glycoproteins generally are either O-linked (to lhreonine or serine) or N-linked (to asparagine). The N-linked glycans extend from the peptide through a constant core region that comprises two 2-acetamido2-deoxyglucosamine (GlcNAc) residues and a branched mannose (Man) trisaccharide. Except for reducing-terminus fucosylation, this pentasaccharide core (Man3GlcNAcx) remains unmodified during biosynthesis for all N-linked glycans. Extensions from this core (Man3GlcNAc~-peptide) gives rise to antenna in three basic motifs (glycotypes); high mannose (see Table I), hybrid, and complex (see Table II). Generally, high-mannose glycans possess only Man, the complex glycans, GlcNAc, and galactose (Gal), while the hybrid antennae include residues of both glycotypes. Galactose residues frequently terminate with neuraminic acid (Neu5Ac), a feature termed capping. These basic components make up most glycoproteins, with annotations to these structures by appropriately positioned fucose, sulfate, phosphate, and acyl groups giving rise to numerous specific biological activities. Glycan composition provides an indication of glycotype and ES of an intact glycoprotein details the glycoform distribution. These data provide a basis for directing additional structural inquiry. Although not always experimentally possible, intact profiles would best represent glycan heterogeneity, uncompromised by selective enzymatic and isolation techniques. Deglycosylation may be discriminating or incomplete and chromatography could enrich or exclude particular subforms, thus a constant challenge during '~B. B. Reinholdand V. N. Reinhold,.I. Am. Soc. Mass Spectrom. 3, 207 (1992).

382

[ 161

MASS SPECTROMETRY

TABLE I LIST OF THEORETICAL MANNOSIDASE TRIMMING PR()I)U('IS "FItA'I MAY BE ANTICIPATED FROM HI(;H-MANNOSE GLY('OTYI'E WlrlllN RANGE Man5 ,~GIcNAc" Product

Struclure

2M6 9-26-A8C 2 "M 6 MTM~ ~-~31cNAc M~--.--M/3

M,~

Ms

M7

8-24-BC M2--M6 M2-M~ M~V~.GIcNAc M-M/3

8-24-AC MLM 4 _ M~~"M'6 ~M~--GIcNAc MZM ~-./3

-MTM 2 .M.~M.6 ~ ~M4--GIcNAc M-M~M/3

7-22-A M~M.6 ^ M;~ "M4GIcNAc M3-M£M4

7-22-B M6 2->M~6 M--M2 M4GIcNAc M-M/3

7-22-C M2--M~ M~M'~44431cNAc M--M/3

7-18-AB M2M ~M'6 4 "M-GIcNAc M2_M2_M4

M2M 6 7-21-AC '~M.6 ~ 2 "M4GIcNAc M-~M--M/3

M2_M6 7-22-BC ..2. ~_M.6 M--M7 ",M4_.GIcNAc M/3

6-16-A

M~

6-19-A M6 ~'M.6 2 ~M4-GIcNAc

2 M~ M'~M4_GIcNAc

.-.-44 .a.4" .

.~-.-.4 G-20-c

M3/"'c~@_GIcNA M / 3

~";~..6

M--M7 M/3"M~GIcNAc

M

M5

2 _~ M.6

~-I~-A

6 . M~@-GIcNAc ML_M':-M/3

6-20 M~M.6 M~" "~44GIcNAc

M~.4

.:.4.4

~-]9-C

M2_M~M4--GIcNAc

6-1 6-B

-MM;" 2_MXM4_GIcNAc

>18

M7 ",M4_GIcNAc M/3

G-20-a

8-24-AB

s-]r-c

M4_GIcNAc M/3 M6M 6"

5-17

M 2 M,,,M4_GIcNA c ~ ' -

s-14-B

M M3/McM~GIcNA . 6 M/3 _5_-14

. }M\6 M 2 M~IcNAc M_M/3

"Three-digit notation (e.g., 9-26-ABC) represents mannose residues, sum of linkages, and antenna identification: A, bottom: B, middle; and C, top.

[1 6]

CARBOHYDRAFE SEQUENCE A N A L Y S I S

BY E S - M S

383

T A B L E 11 PARIIAI LIS I OF COMI'LL:X GLY(()I'YPI2 a l Y('ANS COMPII.tH) Wllll MOLE('ULAI-t W[I(Ht-I AND PARI!NT IONS IN +2, +3, AND + 4 (THAP,(iE S I A I ' I AS S()I-)II!M ADDU('TSa M,

[MNae]/2

[MNa3]/3

[MNaa]/4

BiNA~ BiN A ,

2583.8 2945.2

1314.9 ~2 1495.6 ' 2

0884.3 3 1004.7 ~

0669.0 4 0759.0 4

BiLac/NA/

3033.3

1539.7' ?

1034.1 ~

0781.3 4

BiLac.NAt BiLaqNAe BiLac~NA,

3482.8 3394.7 3844.2

1764.4 ~2 1720.4 ~2 1945.1 2

1183.9 ~ 1154.6 3 13114.4 3

0893.7 4 0871.7 4 1/984.1 4

TriNAI

3033.3

1539.7 "

1034.4 ~

0781.3 4

TriNA, TriNA~ "l'riLaqNAi

3394.7 3756.1 3482.8

1720.4 " 1901.0 " 1764.4 "

1154.6 ~ 1275.0 3 1183.9 ;

11871.7 4 0962.0 4 1/893.7 4

TriLac,NAi Tri l,ac~N A i TriLaqNA2

3932.3 4381.8 3844.2

1989.1 ~ 2213.9 2 1945.1 2

1333.8 ~ 1483.6 -3 1304.4 ~

TriLac,N A, TriLac~NAe T r i L a q NA3 TriLac,NA3

4293.7 4743.2 42115.6 4655.1

2169.8 2 2394.6 3 2125.8-" 2350.5 2

1454.2 1604.1 1424.9 1574.7

TriLac3NA3 TetraNA~

5104.6 3482.8

2575.3 -2 1764.4 2

1724.5 ~ 1183.9 ~

1006.1 I 118.4 0984.1 1096.4 1208.8 1074.4 1186.8 1299.1 0893.7 11984.1 1074.4 1164.7 1186.8 1277.1 1299.1 1389.5 1411.5 1501.9

Glycoform

; ~ ~ ~

T e t r a N A2

3844.2

1945.1 ~-

1304.4 ~

TelraNA.~ T e t r a N A4 TctraLaclNA3 TelraLaclNA4 Tetral,aceNA~ TetraLac~NA4 T e t r a L a c 3 N A3

4205.6 4567.0 4655. I 5{116.5 5104.6 5466.0 5554.1

2125.8" 2306.5 ~ 2350.5 e 2531.2 2 2575.3 2 2756.0 3 2800.1 2

1424.9 1545.3 1574.7 1695.2 1724.5 1845.0 1874.4

TetraLac3NA4

5915.5

2980.8 ~

1994.8 ~

" All listed m o l e c u l a r w e i g h t s includc f u c o s y l a t e d

~ ~ ~ 3 ; 3 3

4 4 4 4 4 4 4 4 4 4 4 4 4 4 a 4 a

and permcthylated structureswith single cation adduction (23 a m u ) ; [M, +

and n m l t i p l c c h a r g i n g a c o n s e q u e n c e of s o d i u m ~ ( 2 3 ) ] / Z is the isotopically m c a s u r c d mass.

isolation is to maintain the fidelity of structures inherent with the starting material. Structural details of attachment site, linkage, and branching are not available in a profile analysis. These features require glycopeptide maps, deglycosylation, and alternative approaches. In this brief chapter we specifically focus on the analysis of isolated glycans and oligosaccharides that are chemically modified to impart greater molecular specificity. Electrospray of these modified analytcs provides marked improvements in sensi-

384

MASSSPECTROMETRY

[ 161

tivity and the products are suitable to CID for an identification of linkage and branchingJ °'11

Oligosaccharide Methylation and Sensitivi O, Methylation can be effectively carried out by dissolving the samples in a suspension of dimethyl sulfoxide (DMSO)-NaOH, (prepared by vortexing DMSO and powdered sodium hydroxide) followed by the addition of methyl iodide. ~2 The methylated glycans are partitioned into chloroform and extraneous materials back extracted with dilute acetic acid. Methylated oligosaccharides protonate poorly and these samples must be infused into the MS with dilute salt (sodium acetate) solutions, the adducting ions of which provide the vehicle for focusing following desolvation. ~LL~Analyzed in this way, several general features are brought together: (1) easy cleanup by organic solvent extraction, (2) a single monodispersed adducting species appropriate for quantitative profiling, (3) following CID and glycosidic cleavage, a measure of sequence and branching, and from secondary crossring cleavages an identification of interresidue linkage, and (4) compatibility with periodate chemistry to augment linkage and branching detail. An indication of detecting sensitivity can be seen in Fig. 1A-C for a commercially available heptasaccharide prepared by methylation. The plots were prepared from scans over a 10-mass unit interval that included the parent ion for a series of sample dilutions. Although all plots were normalized, an indication of sensitivity can be seen with the change in signal-tonoise ratios on dilution. Importantly, these values exhibit a linear response, (Fig. 1D), a feature not observed with matrix desorption techniques, where surface activity and matrix partitioning are complications of fundamental significance. As discussed below, this dynamic range of detection provides for a wide cross-section of structural features to be measured within a single experiment, from the facile glycosidic linkages to the less abundant crossring cleavages important for interresidue linkage assignment. As noted above, carbohydrate samples and their conjugates are heterogeneous. A comparative profile of this distribution is a critical piece of 10 V. N. Reinhold, B. B. Reinhold, and S. Chan, in "Biological Mass Spectrometry: Present and Future" (T. Matsuo, Y. Seyama. R. C. Caprioli, and M. L. Gross, eds.), pp. 403-434. John Wiley & Sons, New York, 1994. it B. B. Reinhold, S.-Y. Chan, L. Reuber, G. C. Walker, and V. N. Reinhold, J. Bacteriol. 176(7), 1997 (1994). t2 I. Ciucanu and F. Kerek, Carbohydr. Res. 131, 209 (1984). L~M. A. Recny, M. A. Luther, M. H. Knoppers, E. A. Neidhardt. S. S. Khandekar, M. F. Concino, P. A. Schimke, U. Moebius, B. B. Reinhold, V. N. Reinhold, and E. L. Reinherz, J. Biol. Chem. 267~ 22428 (1992).

[161

CARBOHYDRATE SEQUENCE ANALYSIS BY E S - M S

385

100, C

1001 B

50' 754 756 "/58 760 762 764 766 768

0

m/z

?A

ol

. . . . . . . . 754 756 758 760 762 764 766 768 m/z

•

754 756 758 760 762 "764766 768 m/z

de"

I

4,," lO !

1o 8

,io I

lO 2

1o 3

Femtomoles

Fl(;. 1. Detecting sensitivity of methylated maltohcptulose using ES MS. Standard sample weighed, melhylated, and analyzed in serial dilutions by direcl infusion: (A) 1.0 pmol; (B) 100 fmok (C) 10 fmol. Scanned over molecular weight-related ion (m/z 759 762) for parent ion, m/z 760.7, [(Glu)7]- 2Na2': (D) plot of ion current vs delccting sensitivity.

analytical information that mirrors three fundamental characteristics: the complement of glycosyltransferases expressed during biosynthesis, the in vivo physiological milieu of the cell, and the time frame of biosynthesis. Production and marketing of recombinant glycoprotein pharmaceuticals have reinforced the need for reliable procedures for product evaluation, and ES-MS can provide a powerful method by which to fingerprint oligomer distributions and monitor glycosylated products with exceptional fidelity. The combination of methylation, sodium adduction, and ES-MS yields a quantitative oligosaccharide profile at excellent sensitivity. These attributes also provide for multiple charging and high mass analysis. As seen in Fig. 2, ES-MS at the upper mass limit of a maltodextran sample indicates a degree of polymerization extending beyond 40 residues. Three ion envelopes, centered at 1108.8 =7, 1289.9 ~, and 1543.3 ~5, represent the oligosaccharides (Glu),, (n = 36-43). The number of charges on each ion can be determined by deconvolution or from the mass intervals between each ion, (e.g., 29.2, 34.0, and 40.8 Da). Thus, 8-kDa oligosaccharides appear to electrospray as expected, and assuming a comparable charge density one may expect 18- to 20-kDa samples to be easily measured.

N-Linked Glycans High-Mannose Glycotype. Deglycosylation techniques are well established either by enzymatic or chemical means, 1 and as discussed above,

386

MASS SPECTROMETRY

1289_9 34.0 6+ 1255.9 1323_9 F-

I00

x

6 =

[161 204_2 5-1-

80 7+

(GIc) 37

4o_8 x 5= zo4_a

6029.2

x

204.2

7 =

I 1543.3 , I 1357_9 1502.4 I I V 7 [ 1584_1

1221_9 ""1

40-

Iil 14 -61 I r I,h11 1-9 k I 1 I t6 5_0

20-

. . . . . . . . . . . . . . .

1100

, .........

1200

, .........

1300

, .........

1400

, ....

, . . . . . . . .

1500

|

1600

FIG. 2. ES-MS of methylated maltodextran sample [degree of polymerization (DP), 2-43] analyzed by direct infusion; spectrum taken at the high mass end, m/z 1543.,3 = [(Glc),~] • 5Na s-. Mass intervals determine degree of charging (e.g., 204.2/34 = 6).

glycan m e t h y l a t i o n a n d E S - M S p r o v i d e sensitivity into the l o w - p i c o m o l e range. A s an e x a m p l e of E S - M S profiling, glycans w e r e o b t a i n e d from r i b o n u c l e a s e B (single-site N - l i n k e d h i g h - m a n n o s e g l y c o t y p e ) following e n z y m a t i c d e g l y c o s y l a t i o n , m e t h y l a t i o m e x t r a c t i o n , a n d E S - M S (Fig. 3). Mass intervals of 162.2/2 indicate h e x o s e residues, a f e a t u r e c h a r a c t e r i s t i c

100

678.9 MansGN +2

80

+2

1.)

N-Glycanase

2.)

Methylation

3.)

ES-MS

Man6GN

60

40

0.1 n m o l e R N a s e B

781.£

Man4GN+2

5~6,.7

400

6oo

Man7GN÷2

/~955.a Man~GN +2 88a./I / ,~ I I [/MangGN

+1

soo

16'oo 18oo

1ooo

lioo

14oo

FIG. 3. ES-MS profile analysis of high-mannose glycans obtained from ribonuclease B by cndoglycosidasc digestion, direct methylation, extraction, and ES MS. Spectrum showing single charge state for glycan series Man,,GlcNAc (n 4-9). See Scheme VII for detailed structure of MansGlcNAc.

[16]

CARBOHYDRATE SEQUENCE ANALYSIS BY E S - M S

387

of a high-mannose glycotype. Glycomers were detected in the two charge states, m/z 576.7, 678.9, 781.0, 883.1, 985.3, and 1087.6 representing the distribution, (Man),GlcNAc (n - 4-9). When compared with capillary electrophoresis of the intact glycoprotein (inset, Fig. 3), these spectra showed an identical profile, suggesting that deglycosylation, methylation, and ES-MS exhibited no disproportionate influence. Some carbohydrate glycans, however, are isomeric and molecular weight profiles fail to resolve individual glycomers. For this information, CID or further chemical modification must be considered (see below). Complex Glycotype. For glycans of the complex type mass profiling provides an assessment of neuraminyl capping and fucosylation, and partial insight into antenna extension and branching, heterogeneity frequently encountered in this glycotype. Presented in Fig. 4 is the glycan profile obtained from a single N-linked site following deglycosylation and methylation. Ion mass intervals equal to Neu5Ac and lactosylamine (Gall 4GlcNAc) indicate the glycans to be of the complex type. The most abundant ion, m/z 1544.6, can be accounted for as a fucosylated TetraNA4 glycan in the three-charged state (Table II, m/z 1545.3--~), with higher homologs of one and two additional lactosamine residues at m/z 1694.8 ~ and 1844.4 -3. A comparable series can be observed at m/z 1424.3 -3, 1574.3-3 3+

TetraNA

100 -

80-

60-

40-

544.6 TetraNA3NgNA3+ / TetraNA44+ TriNA2NgNA3+ / Te~ LacNA43+ / TetraNA3+ / / TetraLacNA33+/ / / 2 / // ~ TetraLac2NA33+ TriNA34+ ~ TriNA33+ / 3 / / ~ // / \ \ ' TetraNA3* / / ~ / BiNA23+ ~L ~ \ /// / / 1694.8 / TriN/A32+ \ / \XriNA~+ \ \2

\

\ \ / 1424.3 // ~ 1274.8 / I BiNA~+ // 11646 ' // [ ] 1574.3

',

l

936.5\ /~ [ 10046 I ~ I

I |[2~

14:ool

0

--~l |

TriNA2* / / ' / / Totr~Lac:A?+ / / 1900.0 / I 7 7'19.8/ i / ./ 1 / 18~4 4 X ,~ i , | 1689.6 1724.6 l / I TetraNA2"-

20I

1000

I

1200

I

J

1400

1600

I

1800

I

2000

m/z I:'1(;. 4. ES MS profile analysis of complex glycans obtained from erythropoietin glycopeptide (N-83) following N-glycanase deglycosylation, methylation, and ES-MS. Spectrum showing both +2, +3, and +4 charge states. Bi-, Tri-, and Tetra-, number of anlennae: NA, neuraminic acid: kac, Gal(l ~ 4)GIcNAc. See Scheme X for detailed structure of TriNA3.

388

MASSSPECTROMETRY

[ 161

and 1724.6-3, corresponding to the trineuraminyl capped glycans. From this single N-linked site, 11 major structures were detected, all fully fucosylated and never showing more than 2 uncapped antenna (TetraNA:), or 3 additional polylactosamine residues. From ion abundance this profile provides a direct measure of glycan concentration and characterizes its composite glycoform distribution. The task of understanding ion profiles of this complexity is greatly simplified by use of composition-mass tables compiled for each homologous series, (e.g., Neu5Ac~, Neu5Ac2, Neu5Ac3, Lact, Lac2, Lac3), at +2, +3, and +4 charge states (natural abundance) (Table I1). Linkage and Branching Inlormation Collision-Induced Dissociation (ES-MS/CID/MS) In addition to enhanced ionization efficiency and easier product isolation, methylation stabilizes monomers from noninformative, small mass losses following CID. Probably the most significant attributes are a series of glycosidic and cross-ring cleavages that combine to allow an understanding of sequence, branching and linkage. Three factors make this possible: first, the adducting ion(s) show a statistical distribution for all residues in the polymer, and thus, for any glycosidic bond ruptured, the probability is high that both product ions will appear charged and mass measurable. Second, the glycosidic cleavage is unsymmetrical, e.g., nonreducing fragments eliminate across the C-1-C-2 bond to become unsaturated, while the reducing end fragment retains the glycosidic oxygen and accepts a hydrogen (Scheme II). Thus, reducing end and nonreducing fragments show a respective 14- and 32-Da decrement from their methylated residue weights, providing a direct indication of Reducing Terminus

.°o--~

R

~-0

M°o~

O~ Me

-32 IMe-(OGIc-H) n ]Na+

R...~.~

Non-reducing Terminus

S{'ln~MI 1I

[16]

CARBOHYDRATE SEOUEN('E ANALYSIS BY ES-MS

389

origin. This feature is not only helpful in defining each fragment, but is enormously important in the structural interpretation of branched molecules, (see below). The mass disparity between these fragments could not be assessed in the absence of methylation. Third, the mass of residual crossring fragments pendent to nonreducing termini allows a direct assignment of linkage type. Collision-Induced Dissociation of Linear Oligosaccharide. The significance of these features can be best illustrated by considering the CID spectrum of a methylated, linear homopolymer represented by the notation CH,~(OGIc)7-OCH~ (Fig. 5). Collision of the parent ion, m/z 1498, [CH~(OGIc)v-OCHd • Na +, provided six abundant glycosidic fragments starting at m/z 259 and ending with m/z 1280. The mass of these ions is consistent with reducing end fragments, [H-(OGlc),,-OCH3]. Na- (n 1-6), exhibiting the unsymmetrical rupture with increments of one hexosyl residue. An additional six nonreducing fragments can be observed starting at m/z 241 and ending with m/z 1262, [CH3(OGlc),,-(OGlc-H)] • Na + (n 0-5) (see inset, Fig. 5). The two major ion series define sequence with unique masses from both the reducing and nonreducing termini. It is interesting to note from their abundance that metal ion adduction shows greater tenacity

241 445 650 854 1058 1262 1280 1t}76 872 668

463 259

668

872

463

1280

I°° 259 241

200

- MM-

[

440

- MM~

854 -MMMM

1076

1

1058

X ] i - MMMMM~-"

i.l. l ...... 11 _1.1 ..... J. I I 680 92{J

1262 -Mc()H~

,..L

. . . . . .

I

1160

I

1400

FI~;. 5. Tandem MS; collision-induccd dissociation (CID) spectrum of ES ionized methylated maltoheptulose, m/z 760.7, [(Glc)7]-2Na 2. Major fragments as a result of glycosidic cleawigc from reducing and nonrcducing ends.

390

[ 161

MASSS P E C T R O M E T R Y Man 2 Man,,xs a Man _ Me n ~ ; 3 "~lan _GN Man 2Man,X%_10e7

Mane

b

Man/eMane "~Man--GN Man2 M a n 2 M a ~ 1 1 0 1

SCHEME Ill

to reducing end fragments than nonreducing termini (cf. Fig. 5, mlz 872 and 650). Collision-Induced Dissociation of Branched Structures. Fragments derived from a glycosidic rupture can undergo additional cleavage(s) at alternate sites that define branching. Such fragments are diminished in abundance, but are easily detected from the low ES background. With methylated samples each glycosidic rupture provides a mass shift that identifies branching frequency. This important difference allows a unique measure of branching not available with underivatized samples and allows selected structural elements to be readily detected, even within structural isomers. As an example of this, a Man7GlcNAc structure responsible for CD2-LFA-3 cell adhesion was isolated and provided a CID spectrum indicating a trimannosyl loss fragment, m/z 1087. This was characterized as the double-cleavage ion and was accounted for by considering the loss of three mannosyl residues from two loci (i.e., two glycosidic cleavages), because it was 28 Da (2 × 14) lower than the fully methylated ion (Scheme Ilia). lx14 By contrast, a Man7GlcNAc glycan isolated from ribonuclease B showed a different CID spectrum, also with a unique trimannosyl loss fragment 14 Da higher, m/z 1101. This indicated a single glycosidic rupture that could only arise from a completed lower antenna (Scheme IIIb). Thus, the CD2 glycomer was either 7-22-B or 7-22-C (Table I), which was resolved by further examination of cross-ring cleavages (see below) to be the latter structure. Collision-Induced Dissociation and Interresidue Linkage. Several laboratories have presented data indicating CID spectra could yield linkage information. Reports have utilized differential ion abundance of glycosidic fragments] 5,1~' unique ions generated by periodate oxidation, :~') lithium 14 B. B. Reinhold, M. A. Recny, M. H. Knoppers, E. L. Reinherz. and V. N. Reinhold. in "Techniques in Protein Chemistry 1II" (R. H. Angeletti, ed.), p. 287. Academic Press, New York, 1992. t5 R. A. Laine, K. M. Pamidimukkala, A. D. French, R. W. Hall, S. A. Abbas, R. K. Jain. and K. L. Maria, J. Am. Chem. Soc. 110, 6931 (1988). t~ R. A. Laine, E. Yoon, T. J. Mahier, S. A. Abbas, B. de lappc, R. K. Jain, and K. L. Matte. Biol. Mass Spectrom. 20, 5(15 ( 1991). 17 A. S. AngcI, F. Lindh, and B. Nilsson. Carbohydr. Res. 168, 15 (1987). ts A. S. AngcI and B. Nilsson, Biomed. Environ. Mass Spectrom. 19, 72I (1990). t~ A. S. Angel, P. Lipniunas, K. Erlansson, and B. Nilsson, Carbohydr. Res. 221, 17 (1991).

[16]

CARBOHYDRATE SEQUENCE ANALYSIS BY E S - M S

391

q B ~

~OCHs

S('ImM~ IV

ion chelation, > or high-energy CID of methylated samples. 21 This latter report continued studies, initiated three decades earlier, showing specific fragments for each of the three linkage types (1 ---+ 6, 1 --+ 4, and 1 --~ 2) using methylated disaccharides fragmented by electron impact. ~2 In this study, two sets of ions carried linkage information and both formed as a result of proximal ring rupture with residual fragments attached through the intervening glycosidic bond (Scheme IV). As expected, the two bond cross-ring ruptures of the A ring were found to be common among all linkage types, while rupture of the B ring provided ions reflecting linkage position (Scheme IV). Because only high-energy electron ionization was available, larger saccharides could not be evaluated for a higher oligomer sequence strategy. With the development of "soft" ionization techniques, high molecular weight polymers could be ionized and much attention was directed to native (unmodified) oligosaccharides, where minor cross-ring cleavages were not detected. The differential instrumental approach of low energy for ionization with the higher energies needed for detailed fragmentation was not available. This was partly resolved when two groups 2324 independently demonstrated that ions could be individually fragmented by introducing a collision gas between the magnet and electrostatic analyzer, providing a high-energy collision-induced dissociation (CID). Most importantly, this new technique provided spectra qualitatively similar to those obtained by electron ionization. Thus, instruments combining "soft" ionization and CID exposed for the first time the possibility of low-energy polymer ionization with the higher energy needs of sequence. As mentioned above, highenergy CID of methylated oligosaccharides ionized by FAB provided crossring fragments that identify linkage position in oligomers of 5-10 residues, .,o Z. Zhou. O. Ogden. and J. Lcary..I. Org. Chem. 55, 5444 (1990). 21 j. Lemoine, B. Fournet, D. Despeyroux. K. R. Jennings, R. Rosenberg, and E. dc I loffmann, J. Am. Soc. Mass Spectrom. 4, 197 (1993). 2~ O. S. Chizhov, L. A. Polyakova, and N. K. Kochetkov. Dokladv Akademii Nauk SSSR lSS, 685 (1964). > K. R. Jennings, Int. Y. Mass Spectrom. h m Phys. 1, 227 (1968). 24 W. F. Hadden and F. W. McLafferly, ,L Am. Chem. Soc. 90, 4745 (1968).

392

MASSSPECTROMETRY

[16]

while collisions at lower energies were reported to be unsuccessful. 21 The ballistic technique of FAB requires a desorption matrix that contributes large backgrounds. This is not a complication for the detection of labile peptide or glycosidic bonds, but does constrain detailed investigation of minor fragments needed to understand oligosaccharide linkage and branching. Electrospray ionization, s by contrast, is a desolvation technique not compromised by a matrix and provides a better opportunity to evaluate minor fragments. We reevaluated the earlier work of Chizhov et a t , 22 and applied it to several structural problems using ES and the lower CID energies available in the triple quadrupole. 2526 The spectral differences in CID between the high-energy sector and the low-energy quadrupole instruments suggest that fragmentation pathways in the former instrument are less selective. In high-energy collisions, energy is deposited on the ion in a single collisional event, in a short time frame. In the triple quadrupole, by contrast, collisions occur over much longer time scales, allowing multiple events. In these circumstances dissociation paths having low activation energies will predominate even if their associated phase space volume or entropy is low. Thus, the direct correlation of abundance to collision energy may not follow, and this could account for the qualitative differences observed in high- and low-energy CID spectra. As indicated above, a linkage between any two residues can be determined by ions defining an intact glycosidic bond linked to fragments from the reducing side ring, thus only nonreducing terminal fragments will carry linkage information. Fortunately, each linkage type gives rise to a different pendant fragment that defines interresidue linkage, e.g., 2-0 linkages, a 74-Da increment; 3-0 linkages, no related ions; 4-0 linkages, an 88-Da increment; and 6-0 linkages, a combination of two ions, 60- and 88-Da increments (Scheme V). Thus, for the simple (1 ~ 4)-linked maltoheptulose (Fig. 5), each nonreducing glycosidic fragment can be identified at m/z 241, 445, 650, 854, 1058, and 1262 (inset, Fig. 5). Because linkage-defining ions represent mass increments, they are found as satellite peaks to the right of these glycosidic fragments. To appreciate these structural details, an expansion of the disaccharide region, m/z 380-580, is presented in Fig. 6. in this spectral expansion, three cross-ring cleavage ions can be identified, m/z 491,519, and 533, but only the latter two ions provide linkage information. The absence of the rn/z 505 ion (+60), with the detection of m/z 533

25 M. A. Velardo, R. A. Bretthauer, A. Boutaud, B. B. Reinhold, V. N. Reinhold, and F. J. Castellino, J. Biol. Chem. 256, 17902 (1993). 2~ M. A. Deeg, D. R. H m n p h r e y , S. H. Yang, T. R. Ferguson, V. N. Reinhold, and T. R. Rosenberry, J. Biol. Chem. 267, 18573 (1992).

[16]

CARBOHYDRATE SEQUENCE ANALYSISBY ES-MS 6

K ~.~ Non-rc~t~

O +88~1

/

+*++.-" ~

-

~

393

+60~1~

.++r,++++

I

i ~

o~

!) +74"tJ

Reducing-end

S('ttI MI! V

and 519 (methyl loss fragment), indicates the linkage to be 4-1inked. The former ion, m/z 491, represents a reducing end fragment that carries no linkage information. Because this sample was a homopolymer, all fragment intervals are identically shifted only by 204 Da, the residue mass of each monomer.

~__

0

q )--0

463 ~

~ t

:~-0

/.-o, UO~

~

9 / - o , 4 4 .~ ,~-o

~-O~

/ ~ - O ~ :'

/O-

445 533 431 519

463 491

431

~o

~o

44

463 491

415

519 445

a 380

420

,uL a I

460

500

.L

, L .L, I

540

I

5gO

Fl(;. 6. Tandem MS; collision-induced dissociation (CID) spectrum of ES ionized methylatcd malloheptulosc, m/z 760.7, [(GlcT)].2Na e'. Expanded mass range from Fig. 5 (m/= 380 580), detailing cross-ring cleavaoes

394

MASS S P E C F R O M E T R Y

[ 16]

In addition to an unsymmetrical cleavage at the glycosidic bond, nonreducing and reducing fragments from N-linked high-mannose samples are easy to define by their additional lack of molecular symmetry. With internal or double-cleavage fragments, however, where this symmetry is lost, the glycosidic rupture assists in defining polarity. Establishing the nonreducing polarity of an ion is important, for it defines the area to search for linkage fragments. Linkage discrimination between the high-mannose glycans often requires an examination of the cross-ring cleavages associated with 6- and 3-1inked antenna. In theory, when provided any individual structure within the high-mannose motif, a combination of glycosidic fragments and crossring cleavages will provide adequate linkage, sequence, and branching information to assign correctly each of the 26 high-mannose glycomers, Man,,GlcNAc (n - 4-9) (Table I). As an example of how these principles can define structural detail, the analysis of a single MansGlcNAc glycomer is described below. The composition, MansGlcNAc, could fit any of three possible structures of the high-mannose glycotype (8-24-AB, 8-24-AC, and 8-24-BC: Table I). The doubly charged molecular weight-related ion was selected, m/,, 978.2 .2 [(Man)sGlcNAc]. 2Na :~, for CID, which provided the spectrum in Fig. 7. Single glycosidic cleavage fragments dominate the spectrum, with a smaller 1525.0

985.4

445.3

1947.8

1320.9

876.3

649.4

1730.1

~ 16.9

853.4 1

I

I

I

200

560

92O

aV'z

1280

1.,..

I

~., [ t, I

I

1640

2000

Fl~;. 7. T a n d e m MS; c o l l i s i o n - i n d u c e d d i s s o c i a t i o n ( C I D ) s p e c t r u m of ES i o n i z e d [ ( M a n ) s G l c N A c ] - 2 N a e' , tn/z 978.2 2.

[16]

CARBOtlYDRATE SEOtJENCE ANALYSIS BY E S - M S

C

B A

Man ~17~

s.~

%=-'G,cN,o

Man~M~Man~' i

~,,,:

,,~3

395

15 .o 13 .9

Man aMan ,z \SManAGlcNAc Man ~Man --2Man/S

" "

8-24-AB

8-24-AC S('lll MI! V I

set of double glycosidic and cross-ring cleavages. The highest mass ion can be accounted for by sodium loss, m/z 1947.8 [(Man)~GlcNAc]. Na ~, followed by three reducing end fragments, all indicating a single glycosidic cleavage (e.g., 14 Da lower than their methylated mass composition), m/z 1730.1, [(Man)vGlcNAc] • Na-; m/z 1525.0, [(Man)(,GlcNAc] - Na-: and m/~ 1320.9, [(Man)sGlcNAc].Na + (Scheme VI). Only the latter fragment defines a specific segment, loss of the completed " A " antenna. This is supported by the corresponding nonreducing fragments, m/z 445.3, [CH3(OGIc)-(OGlc-H)]" N a , and m/z 649.4, [CH3(OGlc)2-(OGlc-H)]N a , with the latter ion again defining a completed 3-O-linked " A " antenna. These data eliminate the 8-24-BC glycomer, leaving the two structures in Scheme V1 as possible candidates. Defining which of these two structures are present (8-24-AC, 8-24-AB), requires locating the 2-1inked terminal mannose to either the " B " or "C'" antenna, a problem resolved with cross-ring cleavage fragments. Three nonreducing terminal fragments, all showing 60-Da increments, define a completed "C'" antenna, m/z 505.3,695.6, and 913.7 (Scheme VII). Expansion of Fig. 7 between m/z 640-750 (Fig. 8) and m/z 84/)-980 (Fig. 9) shows two of these latter fragments and their associated 88-Da increment. Equally supportive are the nonreducing fragments, m/z 853.4, [(Man)4] - Na - (which must include components of the " B " and " C " antenna) and the missing ions within each sequence, notably [(Man)3GlcNAc]-Na ", [(Man)~,GlcNAc]. Na +, and (Man)5 • Na ' .

505.3~695.~913.7'~ Ma n 2 M a ( n ~ . (+e~J(+y Man 6~la n~GIcNA c Man2Man~Man/a 8-24-AC SCHEME VII

MASS SPECI'ROMETRY

396

[ 161

Man -- Man-- Man

\

649.4 Man-Man. Man, Man .Man- GIcNAc+2 -Man-Man

/

Man - Man\ 6 ^~H2 /Man-O

.~Man, / Man ,Man - G I c N A c + /

/

~ -

Man / ~Man ] f-. Man "OH|

/

", / / ,Man - - GIcNAc ~ n/ /-Man" ~ 695.6/

t /Man-Man~'°HMan /

66 .e,I I

l

Man - Man \ Man 6-.-!H 2

6~., II

16,2.1 -~0.6

I

640

OCH3 I CH ICIH

/

/r

li

CHO I

/Man

/ "

GlcNAc

II

/

I

I

662

-

I

684

I

706

I

728

750

FI(}. 8. Tandem MS; collision-induced dissociation (CII)) spectrum of ES ionized [(Man)s GIcNAc]-2Na e~, m/= 985.4 <. Expanded mass range (m/z 640 750) from Fig. 7, showing collision product ion detail for obtaining linkage and branching detail.

Man Man Man +Z Man Man GICNAC

-M.~-M~n 876.1 Marl GI~N,kc -Man Man /~

871.6

Man Marl ManOH Man"

853.7

~n Man, BCH2 Man

Man\

Ma~

/Man GIcNAc

......

I

8J~4"5~ 840

868

9H2 Man~0

Man-Mao\ Man"

.....

8

[

~H

Man o

//

/

89 .6 913.7

Man"

[

OCH

941.7

"

~

I

I

896

924

. . . . .

~. . . .

952

98o

FIcl. 9. Tandcm MS: collision-induced dissociation (CID) spcctrum of ES ionizcd [(Man)sGIcNAc]. 2Na >, m/z 985.4 '2. Expanded mass range (m/z 840 980) from Fig. 7, showing product ion detail for obtaining linkage and branching detail.

[16]

CARBOHYI)RATE SEQUENCE

ANAI.YS1SBY ES MS

397

Identifying components within isomeric mixtures can be particularly difficult where the power of ion selection for CID is lost. However, as demonstrated above, any one component may give rise to a unique fragment and these ions can be utilized to unravel mixtures. The assignment of ion structures in the above data has been consistent with CD3 permethylation and deuterium reduction prior to permethylation.

Oxidation, Reduction, and Methylation o]" N-Linked Glycans Introduction. A l t h o u g h C I D analysis provides considerable insight into the complexities of glycan structure, difficulties may still arise with isomeric mixtures. Under these circumstances a quite different approach for understanding structural detail can be utilized. This relies on chemical selectivity to differentiate isomeric structures by modifying residue weights in a manner unique to linkage type (Scheme VIII). This chemical method exploits the presence of adjacent hydroxyl groups and their susceptibility to cleavage by oxidation, w 1,~ The shift in molecular weight on oxidation, reduction, and methylation (ORM) is a sum of residue weights that change with each linkage type (Scheme VIII).

OCH2

OCfL

i

~

191 (2,&linked) {4,6qinkcd)

162 (6-1raked)

MeO

OMe

R,,l~,~,,~O

MeOCH~

MeO~ R,~'O

206 (2 linked) (4-1inked)

R MoO

MeOCH,

Me()CII,

R ~ O

]

OMe

189 (3,6-1inked) (2,3-1inked) (3,4-1inked) (2,4-1inked)

204 (3

O

linked)

177

Me ()Me S{m ME VIII

] ()Me

(Terminal)

398

MASSSPECTROMETRY

[ 161

In the information acquired, ORM parallels methylation analysis (methylated alditol acetates), but the amounts of sample required are considerably less. Moreover, the products can be subjected to further collision analysis for sequence, both linkage and branching. Where tandem MS instrumentation is unavailable, glycan profiling following ORM limits the number of structures for consideration, and in some cases may define specific glycomers. Oxidation, Reduction, and Methylation of High-Mannose Glycans. The following example illustrates the advantages of ORM chemistry with a high-mannose glycan mixture isolated from a plant glycoprotein. The fraction was divided, one fraction methylated with CD3I and the other prepared by ORM chemistry. The CD3-methylated glycans showed four major ions, [(Man)5_sGlcNAc]. 2Na =+, when analyzed by ES-MS (Fig. 10a). From Table I, this combination of 4 ions indicates 22 possible candidates should be considered. The second fraction showed the appropriate mass shift for four

1028.9 100 -

932.7 I00-

830. l 921.9

80"

80

727.0 749.0

815.3 60-

60

40

40-

20

709.0

|

20-

I

645.9

i

I

500

1000 m/z

1

'

I

500

1000 m/z

FI(;. 10. ES MS prolilc analysis of methylalcd glycans obtained from a soybean lcclin prepared by endoglycosidase treatment, methylation, extraction, and direct MS infusion. (a) Parenl ions, m/z 709.0, [(Man)sGlcNAc]" 2 N a 2 : m/z 815.3, [(Man)(,GlcNAc]- 2 N a ~ : m/z 921.9, [(Man)TGlcNAc] • 2Na=': m/z 1028.9, [(Man),~GlcNAc] • 2Na 2+, respectively. (b) Same glycan mixture analysis preceded by periodate oxidation and reduction before methylation: p a r c m ions, m/z 645.9 [(Man)sGlcNAc] • 2Na=; m/z 727.0 [(Man)~,GlcNAc] - 2 N a 2 ; m/= 749.0 [(Man)~GlcNAc] .2Na2+: m/z 830.1 [(Man)TGlcNAc] .2NaZ-: and m/z 932.7, [(Man),~GlcNAc] • 2Na ~ , respectively.

[16]

CARBOHYDRATE

S E Q U E N C E A N A L Y S [ S BY E S - M S

399

of the ions, with an additional ion for [(Man)~,GlcNAc] • 2Na 2+, m/z 727.0 -2 and 749.0 -2 (Fig. 10b). The ion intervals for all fragments indicate a highmannose glycotype and the mass shifts due to ORM chemistry limit the isomeric possibilities to four for the ManvGlcNAc (7-22-A, 7-22-B, 7-22-C, and 7-22-BC), two for the Man~,GlcNAc at m/z 748.9 ~2 (6-16-A and 6-16-B), and three for the Man~,GlcNAc at m/z 727.0 -2 (6-20-B, 6-20-C, and 6-20), decreasing the glycomer possibilities by 10. The chemistry of ORM does not limit the MansGlcNAc possibilities and they remain at three (8-24-AC. 8-24-AB, and 8-24-BC). Presented in Fig. 1 l is the CID spectrum of the MansGlcNAc peak along with an inset of the determined structure compiled from the residue masses. A series of fragments in both the +1 and +2 charge states define the 3-branch (m/z 1666.5, 1461.0, and 1254.5), with this latter ion a unique signature for completion, again limiting the glycomer structure to either 8-24-AB or 8-24-AC. Further identification of this glycomer would require the characterization of cross-ring cleavage fragments as discussed above (Scheme V). Oxidation, Reduction, and Methylation of Complex Glycans. Profiling glycoprotein glycans by ES-MS provides an important first level of structural inquiry relating to glycotype and glycoform distribution. In complex glycans, tl+e more subtle details of structure (e.g., location of polylactosamine groups, neuraminyl capping, linkage, and branching) are positional

933.11

177--206~ • .189

100

177~-.. 80

".

189--

292

177~206~--206" ~ .... 1666.5 1461,0 1254.5

60 844.9 40' 741.5 3!3.0

638.6

1254.5

[ 1078.3

24) '

1461.0 1666.5

i 500

I A

ldOO

,

1842.9 i £..--.

1500 ln/z

Fl(;. 1. Tandem MS and collision-induccd dissociation (CID) spectrum of ES ionized [(Man)sGIcNAc]- 2Na 2~, m/= 933.0 2. sclccled from spectrum presented in Fig. 10b.

400

[ 161

MASS SPECTROMETRY

isomers transparent to profiling (Fig. 4). The chemistry of periodate oxidation provides a selective strategy to unravel these features. Also, the positional location of lactosylamine groups relative to the 4 or 4' core mannose in triantennary structures yields unique products by ORM (Scheme IX). With these structural entities, each neuraminyl residue will decrease the glycan mass by 88 Da, and the 4'-mannosyl moiety will increment by 2 Da (4 Da if a 4'-linked triantenna). Furthermore, a terminal fucosyl group will cause a decrease in mass by 42 Da (adjacent glycols), as will all uncapped galactosyl termini. Not only does the new pattern of ions unravel difficult isomeric problems, it also provides an opportunity to reexamine and verify structures previously profiled following methylation. The mass differences imparted to selected structures following ORM chemistry can be as little as 2 Da (Scheme X), and multiple charging diminishes the measured mass even further, (e.g., 2/z). This can be offset by reducing with NaBD4 during the reduction step. An example of this approach is presented in Fig. 12, which profiles the glycans obtained from one glycopeptide. As an example, the most abundant glycan (m/z 1544.5 -~, TetraNA4), would be expected to undergo a loss of 352 Da from the four neuraminyl groups (C-8 and C-9 loss), 42 Da from the fucosyl residue (C-3 loss), two mass units would be added due to the single c/s-glycol group on

HO-:H

CH,O-CH,

H ~

°

-88 -42

..._ ~

o-cHf CH,O / ~

OH

OCH3

01 4

HO-~ HO-.~H HO-CH ~.

/ L - ~ COOH

oR

NA ct 2.-->3Gal

CH3fHa-OCI~ CHI.OH--

°"

-88

OH Sc'l IIEMH [X

O~

CIL "OCHs COOCHs - -

[161

CARBOHYDRATE SEQUENCE ANALYSIS I3¥ E S - M S

Neuct 2-.>3

G~l,I--~.4 GkN[~ 1 ~ > 2

6

Mtncx I

$

N.,,. 2-.3 G~v->4 GkN~~

Fuc ct1~>6 IoN

4 ~,.,M.,,~

8 7 ..~ N e n ct 2,-->.3 Gall 1 - - > 4 GIcN[3 r ' ~

N e * ¢x 2 - - . 3

401

M a n o.

Gall 1 - - > 4 GleNI3 1 "

262

"

~

6 Man # 1 -->4

GkN ~ 1--~4

GleN

fir N e n c t 2-->.3 Gall1-->4 GIcN~ l-->2MImo. 1

-I7-(, x ~]

~

- 260

SCHEME X

1422.4 TetraNA43+

100TetraNA4. 4

/

TetraNA~ / "+

etraNA3NgNA3+ 80"

TetraLacNA3; 1184.7[ 60-

TetraLaeNA~+

To,r~4+

~ t [

1,1,.,

40-

1572,/

Tetral,ae2NA ~+

t/ X e t r a L a c N < 994.0

'

21;*-

|

/

[

/ TetraLat'2NA4

t/I,

1721.8 //

6O0

800

1000

1200

1400

i 1600

TetraNA3 + ~_-..

•

t 1800

N ,

2000

FI< 12. ES-MS of complex glycans obtained from erythropoictin glycopeptidc (N-83) following N-glycanase treatment, periodate oxidation, reduction, and mcthylatio0.

402

MASSSPECTROMETRY

[161

the 4'-mannose, and the reducing terminus would increment the mass by 16 Da (glucosylaminitol formation during reduction) for a summed loss of 376 Da, rnlz 1419.7 +3, when using NaBH4. Because this sample was reduced with NaBD4, the expected product ion would shift to m/z 1422.9+3. The experimental data provided an abundant product ion at m/z 1422.2 +3 (Fig. 12), accounting for the modifications discussed. The rest of the ions can be rationalized using the same considerations. The significance of this approach was realized in detailing the glycan structures on two adjacent glycopeptides from erythropoietin, N-38 and N-83. The product ions indicated a decided difference in branching (4- vs 4'-) at the two sites with masses at mlz 1181.2 --~ and 1182.5+3, respectively. This was confirmed by reanalysis after mixing the samples and showing resolution of the ion pair. A series of suppositions is frequently made by mass spectroscopists during carbohydrate analysis, that core hexose and aminosugar mass increments, and all anomeric configurations, are as previously assigned. Most of these assumptions are appropriate when assessing modifications to established motifs, but if adequate amounts of material are available these features should be confirmed by nuclear magnetic resonance (NMR). Summary This chapter summarizes several strategies for a more complete understanding of carbohydrate structure with a focus on N-linked glycans. The techniques include periodate oxidation to impart greater molecular specificity, functional group blocking by methylation, electrospray for "soft" and efficient ionization, collision-induced dissociation to obtain detailed fragmentation, and tandem mass spectrometry for mass separation and analysis. The lipophilic products following derivatization contribute to product cleanup by solvent extraction; they also improve sensitivity during ES and, when combined with CID, yield detailed sequence, linkage, and branching information.